Why social proof and conformity works

In 1935 the pioneering social psychologist Mazafer Sherif invited people to take part in an experiment using the autokinetic effect. Participants looked at a point of light in a darkened room and were asked to report whether they thought the light was static or moving, a recreation of a natural phenomenon first observed by astronomers who thought that stars were moving. When participants were asked individually, opinion was equally divided; however, when they were put into groups people tended to agree with the majority, even if this meant contradicting what they’d said originally. Later, when asked individually, they continued to subscribe to the group view. In other words, when placed in the context of a group, people will devalue their own opinion in the interest of developing an arbitrary position that is acceptable to the group.

Consensus is not five opinions merely because five mouths repeat it. Under ambiguity, each answer can contain traces of the answers that came before.

In Sherif’s experiment, a person sits in a dark room watching a fixed point of light. With no reference, it appears to move. “About four inches,” they say. Other participants begin with different answers, then hear one another and converge.

Later, alone, they retain the group estimate.

Nobody necessarily pretended. Under ambiguity, another person’s confidence is information. Perhaps they saw something you missed. The second answer may already include the first, and the third includes both. One observation is copied until it looks like several.

A forecasting group records estimates privately. The range is wide and confidence low. Another group speaks in turn. Its answers cluster quickly around the first confident number. The second meeting feels more productive.

Written estimates make a different conversation possible. The group can ask why one person saw 35 and another saw 80. A hidden assumption about timing emerges. Lose the starting record and that information vanishes during consensus.

Agreement can come from separate evidence, shared evidence or social adjustment. The final number looks identical in each case. Only the record of the starting points preserves the difference.

At the end of Sherif’s experiment, each participant returns to the dark room alone. The point of light is still fixed. Their estimates are closer than they were at the beginning.